Evaluation tests for eye tracking systems.

Area of Science:

• Computer Vision
• Biomedical Engineering
• Ophthalmology

Background:

• Accurate eye movement quantification is crucial for diagnosing neurological and ophthalmological conditions.
• Existing eye tracking algorithms require rigorous performance evaluation.
• Simulating diverse eye image conditions presents a significant challenge.

Purpose of the Study:

• To introduce novel tests for evaluating eye tracking algorithm performance.
• To develop a synthetic eye image generation library for controlled testing.
• To assess the algorithms' ability to quantify vertical and horizontal eye movements.

Main Methods:

• Creation of a specialized library for generating synthetic eye images with controllable parameters.
• Development of a testing framework to evaluate eye tracking algorithms.
• Simulation of various conditions encountered during real-world eye image acquisition.

Main Results:

• The developed tests provide a standardized method for performance evaluation.
• The synthetic image library allows for reproducible and controlled testing scenarios.
• The system effectively quantifies vertical and horizontal eye movements across simulated conditions.

Conclusions:

• The proposed testing methodology and synthetic data generation offer a reliable approach for validating eye tracking algorithms.
• This work facilitates the development of more accurate and robust eye tracking systems.
• The simulation capabilities enhance the assessment of algorithms under diverse and challenging conditions.